Oracle-Based AMM
HyperAMM uses a hybrid oracle system instead of x*y=k to price trades. No bonding curve means no arbitrage-driven impermanent loss.
Why Oracles?
In traditional AMMs, prices come from pool reserves. When the market moves, arbitrageurs extract value from the pool — that's IL. Concentrated liquidity (Uni V3) makes it worse.
HyperAMM sidesteps this by pricing trades from external oracle data. The pool's own reserves don't determine price.
Hybrid Oracle Architecture
Different operations use different oracle sources:
Deposits & Withdrawals: Pyth Pull Oracle
Deposits and withdrawals use Pyth Network pull oracles with:
- Strict staleness checks — Prices must be fresh within a tight age threshold
- Confidence bands — Pyth's confidence intervals protect LPs from price manipulation
Pricing always favors existing LPs:
- Deposits get the less favorable end of the confidence band
- Withdrawals get the less favorable end for the withdrawer
This prevents arb bots from exploiting stale or uncertain prices to drain the pool.
Swaps: HyperCore Prices
Swaps use HyperCore's native price data:
- Perpetual BBO — Real-time order book prices
- Mark price — Fair value from the exchange
- Oracle price — Hyperliquid's aggregate oracle
- Pyth push oracle — For stablecoin pairs (e.g., USDC/USDT0), maintained by the protocol's oracle updater
This enables atomic swap settlement — the swap completes in a single transaction, no waiting for oracle updates.
Swaps need to settle atomically in one tx — waiting for a pull oracle update would make them async. Core's built-in prices are available instantly via L1 state reads. Deposits and withdrawals are already async (executed by keepers next block), so they can use the pull oracle.
Comparison
| x*y=k AMM (Uniswap) | HyperAMM | |
|---|---|---|
| Price source | Pool reserves | External oracles |
| IL risk | High (especially concentrated) | Zero (hedged atomically) |
| Price discovery | Via arbitrage (LPs pay the cost) | Via oracles (no LP cost) |
| Capital efficiency | Moderate (V3 concentrated) | High (oracle-driven) |
| MEV exposure | Sandwich attacks, JIT liquidity | Reduced (oracle-based pricing) |
- Pool Types — NEUTRAL vs BULL mechanics
- Delta Hedging — How trades are hedged atomically
- Dynamic Fees — Fee structure